Measurement Error of Two Three-dimensional Techniques for CBCT Superimposition

Objectives: Objectives: To compare measurement error in the Alberta and Voxel-base techniques for three-dimensional superimposition of CBCT images.
Methods: Methods: CBCT volumetric data was obtained at a six-month interval from nine control patients (not in treatment). A total of eighteen CBCT images were supplied to two independent researchers (one at the University of Alberta and one at the University of North Carolina) for analysis using AVIZO to locate nineteen (19) skeletal landmarks per image. Each group performed their respective methods to superimpose the T1 and T2 images (North Carolina using voxel superimposition, and Alberta using a 6-point transform) to obtain xyz-coordinates in a consistent coordinate system. A paired t-test was used to compare the transformed landmark coordinates at T1 and T2 for twelve landmarks on the facial skeleton.
Results: Results: Both techniques for superimposition of CBCT were successful in aligning the two images (significance ≤ 0.01). North Carolina method showed a slightly higher correlation (r = 0.994 ± 0.005) than then Alberta method (r = 0.923 ± 0.01) in aligning individual coordinates (95% CI).
Conclusions: Conclusions: Both the voxel-base and 6-point coordinate transform were successful in superimposing two CBCT images with high correlation.